The goal of the proposed work is to develop a radically new method for rapid direct mapping of the regions of genetic identity-by-descent between two related individuals, using only DNA samples from the two individuals. The method is intended to allow the simple and powerful but presently impractical "affected relative pair" mapping strategy to be put into practice. The principal advantages of the proposed method are: 1. No other family members need to be examined, and no pedigree information is required. 2. No locus-specific probes are required. 3. The entire genetic map is surveyed at high resolution in a single procedure, rather than through multiple discrete analyses of polymorphic loci at intervals throughout the genetic map. The method takes advantage of the fact that (at least in an outbred Western European or US population) any two allelic sequences that are not inherited from a common recent ancestor have single base differences on average every few hundred (about 300) basepairs. In other words, for an arbitrary pair of relatives, there are around 10-7 potentially informative "sequence polymorphisms" per haploid genome. The proposed method is designed to provide a practical way to exploit this vast source of information for gene mapping. The method involves four steps: 1. Preparing genomic DNA from two related individuals. 2. Preparing and isolating hybrid DNA molecules containing one strand from each individual. 3. Selecting those DNA hybrids that are free of mismatches over many thousands of base-pairs, and therefore very likely to represent sites of genetic identity by descent. 4. Using the resulting large, heterogeneous pool of perfectly, matched DNA hybrids as the template for probes, initially for in situ hybridization to metaphase chromosomes, to identify the map locations of sequences identical by descent between the two tested individuals. With a sufficiently large collection of affected relative pairs, the frequency of genetic concordance at the map position(s) of the trait-determining gene(s) will exceed that expected to occur by chance. In the case of a rare trait uniquely determined by a dominant allele of a single gene (e.g., von Recklinhausen's neurofibromatosis), 10 pairs of relatives sharing the trait would generally be sufficient to map the gene to a single 10 megabase region of the genome. Most of the technical obstacles to implementation of this approach have actually been overcome by previous workers, in other contexts. Thus, while the proposed concatenation and adaptation of available techniques to gene mapping represents a fundamental departure from previous mapping strategies, no fundamental technical breakthroughs are invoked. A closely-related method for mapping rare recessive traits using DNA from only one or a few affected individuals will also be tested.